As vendors withdraw their support for open source infrastructure platform OpenStack, Canonical founder Mark Shuttleworth has said this is a sign of a healthy, maturing project - but that the OpenStack community needs to have a frank conversation with itself about where it’s headed.
Speaking with Computerworld UK, Shuttleworth says he is “excited about the maturing of OpenStack”. “Some people will say it’s going to get boring without all of the fighting, without all of the vendors showing up to have opinions,” he says. “Actually, I think this is what OpenStack needs - maturity.”
In 2016, HP and Cisco both significantly cut their OpenStack operations. Intel pulled the funding on its joint Innovation Centre, originally opened with Rackspace. And Mirantis also took the axe to its OpenStack engineer workforce, following an acquisition of TCP Cloud.
But Shuttleworth says this is a sign that OpenStack is on the right track, crystallising itself as a core infrastructure contender and shedding away gimmicky services that were never going to last - a position he has held for some time, telling Computerworld UK this time last year that the ‘Big Tent’ approach was bound to fail.
“In the enterprise market what we see is customers now coming forward with a much clearer idea of what they really want, which is private infrastructure that behaves like a public cloud,” Shuttleworth says. “They don’t have to worry about it, they don’t have to hire people, they essentially consume it in the same way they consume public infrastructure.” The messaging clearly chimes with Canonical’s offering, which operates OpenStack for clients.
“There’s a lot of angst in the halls because there are vendors who are not going to survive that growing up,” he says. “Those vendors had very colourful pitches about having the best scientists in OpenStack, or the best databases in OpenStack, or whatever their special feature was, but none of that really matters. So for those vendors this maturing is very difficult, you’ve seen HP pull out, you’ve seen Mirantis essentially end of life their installer, Intel and Rackspace.
“But that’s not the same as the end of OpenStack, in fact that’s just the growing up of OpenStack.”
It’s true that OpenStack is maturing - the most recent User Survey, which asked 1,300 respondents from the community to provide feedback - and noted a rise in adoption. The majority of deployments were in 2016 or 2017.
“A good analogy is the dot com bubble,” says Shuttleworth. “Remember 1999, with pets.com? When all that collapsed, the internet didn’t stop. The underlying drivers of cheaper connectivity, faster connectivity, every month, every year, that didn’t change.
“I think it’s kind of the same here. You’re seeing these fantasy projects die… Customers are now starting to appreciate a working OpenStack at predictable economics.”
Shuttleworth is still hammering home his point that OpenStack is most valuable at its core, as infrastructure for VMs on private network. He says that there is a “cultural thing inside OpenStack” that will continue to pose challenges.
“This is a hard thing to say,” he explains. “It’s easy to say there’s a technology problem we have to solve, because everyone can agree that’s fun and interesting. Or to say there’s a competitor we all hate, it’s easy to get people wound up about that. But actually in neither case is that an issue. Amazon isn’t trying to compete with OpenStack - it just isn’t. Amazon is a public cloud and they’re great at it, but they’re not competing with OpenStack.
“Essentially OpenStack’s only competition is itself, and actually the challenge to overcome is how OpenStack talks about OpenStack,” he says.
“It’s never super popular to talk about because you’re really saying the problem isn’t out there, it’s in here. In the same way there was all this hubbub about all these thingies-as-a-service that have now collapsed… we could have avoided that if we had a clearer conversation about really getting the infrastructure right.
“So now I think the real cultural change is for the OpenStack guys to say, for us to say, OK, we are really going to master virtual machines, virtual networks, and virtual disks. We don’t have to worry about containers, because you can do the same Kupernetes on top of Amazon or OpenStack. You can do the same Docker on top of Amazon or Openstack.
“OpenStack doesn’t have to panic about containers, it doesn’t have to add APIs for containers, it doesn’t have to do any of these things - all it has to do is really deliver reliable infrastructure for VMs on demand. That’s the real challenge, to get people here to accept that this is a fantastic mission in and of its own right. It’s not solve all the world’s tech problems, it’s not beat VMware into a pulp, it’s just to get really good at. this crisp economic proposition of VMs on demand on private infrastructure.”
It is an argument Shuttleworth has made for some time. How successful does he think he has been in convincing the wider community? In his opinion, he says, there has been progress “in a sense”.
“I think there are a lot of people now that would say Canonical has consistently called out where OpenStack could focus on the things that really mattered,” he says. “That’s not an ‘I told you so’, that’s an ‘if we want to get better we have got to focus as a community and learn and grow’.”